Hi Tim, looking at your recent optimization of _celt_autocorr, it doesn't appear that the variable "n2" is being used. Cheers, John Ridges